exp/norm: Added some benchmarks for form-specific performance measurements.

R=r
CC=golang-dev
https://golang.org/cl/5605051

src/pkg/exp/norm/normalize_test.go

@@ -495,11 +495,11 @@ func TestAppend(t *testing.T) {
 	runAppendTests(t, "TestString", NFKC, stringF, appendTests)
 }
 
-func doFormBenchmark(b *testing.B, f Form, s string) {
+func doFormBenchmark(b *testing.B, inf, f Form, s string) {
 	b.StopTimer()
-	in := []byte(s)
+	in := inf.Bytes([]byte(s))
 	buf := make([]byte, 2*len(in))
-	b.SetBytes(int64(len(s)))
+	b.SetBytes(int64(len(in)))
 	b.StartTimer()
 	for i := 0; i < b.N; i++ {
 		buf = f.Append(buf[0:0], in...)
@@ -510,16 +510,43 @@ func doFormBenchmark(b *testing.B, f Form, s string) {
 var ascii = strings.Repeat("There is nothing to change here! ", 500)
 
 func BenchmarkNormalizeAsciiNFC(b *testing.B) {
-	doFormBenchmark(b, NFC, ascii)
+	doFormBenchmark(b, NFC, NFC, ascii)
 }
 func BenchmarkNormalizeAsciiNFD(b *testing.B) {
-	doFormBenchmark(b, NFD, ascii)
+	doFormBenchmark(b, NFC, NFD, ascii)
 }
 func BenchmarkNormalizeAsciiNFKC(b *testing.B) {
-	doFormBenchmark(b, NFKC, ascii)
+	doFormBenchmark(b, NFC, NFKC, ascii)
 }
 func BenchmarkNormalizeAsciiNFKD(b *testing.B) {
-	doFormBenchmark(b, NFKD, ascii)
+	doFormBenchmark(b, NFC, NFKD, ascii)
+}
+
+func BenchmarkNormalizeNFC2NFC(b *testing.B) {
+	doFormBenchmark(b, NFC, NFC, txt_all)
+}
+func BenchmarkNormalizeNFC2NFD(b *testing.B) {
+	doFormBenchmark(b, NFC, NFD, txt_all)
+}
+func BenchmarkNormalizeNFD2NFC(b *testing.B) {
+	doFormBenchmark(b, NFD, NFC, txt_all)
+}
+func BenchmarkNormalizeNFD2NFD(b *testing.B) {
+	doFormBenchmark(b, NFD, NFD, txt_all)
+}
+
+// Hangul is often special-cased, so we test it separately.
+func BenchmarkNormalizeHangulNFC2NFC(b *testing.B) {
+	doFormBenchmark(b, NFC, NFC, txt_kr)
+}
+func BenchmarkNormalizeHangulNFC2NFD(b *testing.B) {
+	doFormBenchmark(b, NFC, NFD, txt_kr)
+}
+func BenchmarkNormalizeHangulNFD2NFC(b *testing.B) {
+	doFormBenchmark(b, NFD, NFC, txt_kr)
+}
+func BenchmarkNormalizeHangulNFD2NFD(b *testing.B) {
+	doFormBenchmark(b, NFD, NFD, txt_kr)
 }
 
 func doTextBenchmark(b *testing.B, s string) {
@@ -657,3 +684,6 @@ const txt_cn = `您可以自由： 复制、发行、展览、表演、放映、
 署名 — 您必须按照作者或者许可人指定的方式对作品进行署名。
 相同方式共享 — 如果您改变、转换本作品或者以本作品为基础进行创作，
 您只能采用与本协议相同的许可协议发布基于本作品的演绎作品。`
+
+const txt_cjk = txt_cn + txt_jp + txt_kr
+const txt_all = txt_vn + twoByteUtf8 + threeByteUtf8 + txt_cjk